Photoluminescence response analysis

The PL responses of the samples in this section were plotted using a bi-Gaussian where possible, and using LOWESS (Locally Weighted Scatterplot Smoothing) where necessary.

Lastly, this section feature several measurements at 1 sun laser beam intensity. For these measurements, it was necessary to increase the integration time to 100 seconds, because the perovskite was not luminescent enough to show sufficient intensity counts for an analysis otherwise. The substrates that are used for the experiments in this section consist of glass, 125 nm of ITO, 15 nm of Spiro-TTB, and a varying perovskite layer thickness.

The bandgap of CsPbI2Br is around 1.92 eV, which does generally not depend on the thickness of the perovskite layer. However, for particularly thin layers, the bandgap experiences a slight blue-shift. Figure 4.25 further illustrates this, notably the 70 nm layer blue-shift to 1.93 eVⁱⁱⁱ.

iiiThe 70 nm sample was measured using 1 sun, while the others were measured using 10 suns. In order to make a visual comparison possible, the 10 sun measurements were divided by a factor of 90 for this graph. Furthermore, the detector causes a baseline peak for any 1 sun measurement at 2.03 eV, this should be disregarded.

4.4 Phase segregation analysis 43

Figure 4.25: Initial PL response for different perovskite thicknesses

MAPb(BrXI1−X)3 perovskites segregate completely after being exposed for only a few seconds, as studied by Slotcavage et al [56]. Furthermore, Beal et al. and Zeng et al., in addition to Slotcavage et al., argued that the stoichiometric CsPbI2Br is stable and does not significantly exhibit phase segregation [60, 61, 56]. However, their results are based on short-term experiments and spin-coated perovskite. The long-term CsPbI2Br segregation process for a co-evaporated and stoichiometric compound mixture will be investigated in the analysis of this section. The most significant bandgap shift registered by the experiments was a red-shift from 1.92 eV to 1.82 eV.

For this analysis, PL responses were measured with a laser intensity of 10 suns. The measurement conditions are constant, with the substrates exposed to air, and all measurements performed on the same day. Figure 4.26 shows the PL responses of the investigated substrates, where they were tested at different points in time to investigate their stability. These measurements were done for 165 nm, 185 nm, 230 nm, and 280 nm perovskite layers. From the results, a bandgap red-shift is consistently visible. The quickest segregation happens for the 165 nm perovskite layer, taking only 50 min to reach a bandgap of 1.84 eV. In fact, a second peak in the PL graph appears after just one minute of exposure, signifying the presence of defect energy states with a lower bandgap. The slowest segregation is seen for the 185 nm perovskite substrate. The results are further tabulated in table 4.1.

44 4.4 Phase segregation analysis

Figure 4.26: PL responses measured over time at 10 sun intensity for 165 nm, 185 nm, 230 nm, and 280 nm perovskite layers

4.4 Phase segregation analysis 45

0 min 6 min 30 min 50 min 70 min 100 min 165 nm 1.919 eV 1.903 eV 1.859 eV 1.856 eV - -185 nm 1.920 eV 1.920 eV 1.912 eV 1.903 eV 1.898 eV 1.872 eV 230 nm 1.924 eV 1.912 eV 1.882 eV 1.877 eV 1.877 eV 1.877 eV 280 nm 1.918 eV 1.916 eV 1.892 eV 1.881 eV 1.870 eV 1.848 eV

Table 4.1: PL response bandgap measurements at 10 sun intensity for different times of air exposure for substrates with varying perovskite layer thicknesses

To verify that the bandgap change does not only depend on the air exposure, the PL responses of two samples were compared: one exposed to air, in figure 4.27, and one under a dry, inert N2 atmosphere, in figure 4.28. Both samples have an initial bandgap of 1.917 eV. After 3.4 hours, the substrate exposed to air reached a bandgap of 1.854 eV, while the one in the dry N2 atmosphere reached 1.873 eV in the same time. This implies that the air exposure does have an adverse effect on the substrate photo-stability. The phase segregation is slightly faster in air.

Figure 4.27:

PL response over time under continuous light-exposure of an air-exposed sample

Figure 4.28:

PL response of a sample over time under continuous light-exposure under a dry N2

atmosphere

After this comparison, the N2-exposed substrate was first let to phase segregate to the same bandgap as the air exposed substrate, and then both substrates were subjected to a period of rest in a completely dark environment. For both cells to start at the same bandgap, the sample placed under an N2 atmosphere had to be exposed to the light beam for 44 minutes longer. Figure 4.29 and figure 4.30 illustrate the progression of the

46 4.4 Phase segregation analysis

bandgaps of the substrates left in the dark. Both bandgaps eventually blue-shifted back to a value of 1.907 eV. This further verifies that the red-shift observed under light exposure is mainly caused by the phase segregation process, which is reversible. After one hour in the dark environment, the N2-exposed substrate had shifted back to a bandgap of 1.894 eV. In contrast, the air exposed sample reached a bandgap of 1.873 eV after an hour in the dark. From this data, it is concluded that the air exposed sample takes a longer time to return to its original crystal structure.

Figure 4.29:

PL response after periods of darkness of an air-exposed sample

Figure 4.30:

PL response after periods of darkness of an sample under a dry N2 atmosphere

After this period in the dark, the samples were re-exposed to a laser beam of 10 suns intensity, in order to test their stability after the reversed phase segregation. At this second exposure, the air-exposed sample was already completely phase segregated to a bandgap of 1.825 eV after 80 min, as portrayed in figure 4.31. This is in contrast to the first exposure, where the bandgap had only shifted to 1.854 eV after 3.4 hours. In figure 4.32, the sample in N2 atmosphere exhibits a bandgap shift to 1.845 eV in 80 minutes.

These results show that the phase segregation is significantly accelerated during the second exposure, implying that the precedent exposure weakened or modified the original crystal structure of the perovskite. Furthermore, during the second exposure, the influence of air-exposure became more evident.

4.4 Phase segregation analysis 47

Figure 4.31:

PL response over time under continuous light-exposure of an air-exposed sample at second exposure

Figure 4.32:

PL response of a sample over time under continuous light-exposure under a dry N2

atmosphere at second exposure

To prove that the observed red-shift phenomenon is due to phase segregation and not to degradation, the air-exposed substrate was left to rest in the air-exposed environment for 48 hours, and measured at several points in time to register its degradation. Evaporated layers exhibit conformality issues at the edges of the deposition plate, i.e. a thickness difference is found between the core and edges of the layer. This is attributed to the deposition plate slightly screening the cell from the evaporated material, so that the edges close to it are thinner than the main part of the layer. Consequently, (co-)evaporated perovskite layers usually degrade from the edges of the layer inward, since the edge of the perovskite layer is thinner and thin layers degrade more rapidly, as discussed in section 4.1.2. The edges of the substrate were therefore measured to obtain the most significant degradation results.

48 4.4 Phase segregation analysis

Figure 4.33: PL response of a sample after varying periods of air-exposure in darkness

As portrayed in figure 4.33, after 24 hours the core of the substrate was measured to already observe a slight degradation. From that point, the other measurements were performed at various spots along the edges of the sample. From these measurements, it is evident how the bandgap energy increases due to degradation. This is explained by the fact that the perovskite degrades into its orthorhombic δ-phase, which has a bandgap of 2.85 eV [48], causing a blue-shift in the overall material.

In order to investigate the influence of light intensity of the rate of phase segregation, a sample subjected to a laser beam of 10 suns intensity is compared to a sample subjected to 1 sun intensity. Both samples employ a 190 nm perovskite layer thickness. After 245 minutes, the sample subjected to the 10 suns laser beam attained a bandgap of 1.85 eV, as shown in figure 4.34, while the sample subjected to 1 sun still had a bandgap of 1.89 eV after 245 minutes, though with a second peak starts to appear at a lower bandgap, as shown in figure 4.35 ^iv. However, after 291 minutes, the bandgap change is insignificant and the main bandgap is still at 1.88 eV. It is therefore concluded that the phase segregation happens at a higher rate under the 10 sun light beam. Consequently, this data suggests that the rate of phase transition under light exposure depends on light intensity.

ivThe detector causes a baseline peak in the LOWESS plot for any 1 sun measurement at 2.03 eV, this should be disregarded.

4.4 Phase segregation analysis 49

Figure 4.34:

PL response over time of a sample under continuous light-exposure measured at 10 suns

Figure 4.35:

PL response over time of a sample under continuous light-exposure measured at 1 sun

Also for 1 sun illumination, analysing perovskite thicknesses of 70 nm, 190 nm and 280 nm, it is clear that the time for having phase segregation increases at the increasing of the thickness of the perovskite layer.

The influence of the perovskite layer thickness was further investigated through PL measurements with a laser intensity of one sun. Three samples with different perovskite thicknesses were measured over time, one of 70 nm, one of 190 nm, and one of 280 nm.

For this comparison, the thickness of 70 nm was specifically chosen to verify that the phase segregation occurs more rapidly for thin perovskite layers, in contrast with organic perovskiten [59]. The PL responses are graphed in figure 4.36. As hypothesised, the thin 70 nm perovskite layer exhibits the highest rate of phase segregation, and this rate decreases as the perovskite layer thickness increases. Furthermore, the significant results are tabulated in table 4.2.

50 4.4 Phase segregation analysis

Figure 4.36: PL responses measured over time at 1 sun intensity for 70 nm, 190 nm, and 280 nm perovskite layers

0 min 40 min 80 min 100 min 4 hours 5 hours 70 nm 1.927 eV 1.918 eV 1.841 eV 1.830 eV 1.822 eV -190 nm 1.916 eV 1.916 eV 1.915 eV 1.915 eV 1.886 eV 1.877 eV 280 nm 1.916 eV 1.916 eV 1.916 eV 1.916 eV - 1.907 eV

Table 4.2: PL response bandgap measurements at 1 sun intensity for different times of air exposure for substrates with varying perovskite layer thicknesses

During this analysis, the difference in absorbance between 70 nm and 280 nm perovskite samples was additionally observed. The 70 nm layer absorbs considerably less than the 280 nm layer. This is illustrated by figure 4.37. Furthermore, the difference in reflectivity can be visually confirmed by the appearance of the cells, as the 280 nm sample appears more opaque than the 70 nm sample, which is shown in figure 4.38.

4.4 Phase segregation analysis 51

Figure 4.37: Absorbance comparison of samples of varying perovskite thickness on Spiro-TTB

Figure 4.38: Visual difference in opacity between a 70 nm and a 280 nm perovskite samples

To determine how the phase segregation is influenced by the crystallographic change developed in the perovskite atomic-structure through annealing, three samples were compared after being subjected to varying annealing temperatures. All these samples have a perovskite layer thickness of 190 nm. The first sample is not annealed and it has an initial bandgap of 1.916 eV, the second sample is annealed at 80°C and starts with a bandgap of 1.909 eV, and the last sample is annealed at 100°C and starts with a bandgap of 1.918 eV. Both annealing processes lasted for 10 minutes. These initial bandgaps are graphed in figure 4.39. These initial differences in bandgap might be explained by a crystal structure change that comes with the process of annealing, which will be further investigated by section 4.5.

52 4.4 Phase segregation analysis

Figure 4.39: Initial PL responses for samples after varying annealing steps

The PL responses of the not annealed, 80°C annealed, and 100°C annealed samples are pictured in figure 4.40 ^v. As visible from these plots, only the 100°C annealed sample exhibits a bandgap shift at 72 minutes, the other samples remain insignificantly changed.

From this it can be concluded that annealing at this temperature modifies the perovskite crystal structure in a way that induces a higher rate of phase segregation as compared to annealing at other temperatures, or no annealing at all.

Figure 4.40: PL responses of annealed samples measured over time at 1 sun intensity

A closer comparison is made between the not annealed and the 80°C annealed sample.

vThe detector causes a baseline peak in the LOWESS plot for any 1 sun measurement at 2.03 eV, this should be disregarded.